WHAT’S INSIDE

• Tech News: Stock Highs, USD Lows, Tax Bill, AI Provision, Apple’s "F1" Win

• Company Watch: Crossing Minds, TikTok, Glean, Audos, Campfire

• Feature: Meta's $65B Superintelligence Bet & AI Talent Poaching Spree

• Buzzy Tools: Latest Buzzy tech, AI and financial tools

• Deep Tech: The latest in deep tech, biotech, futurism and more

• Space Tech: Latest news in the space race and aerospace tech

• Crypto: Blockchain and crypto policy and startups or protocols to watch

TECH IN THE NEWS

Market Highs — The S&P 500 and Nasdaq reached record highs, marking their best quarter in a year with double-digit gains. Global stocks soar as Canada drops its digital services tax, boosting US trade talks.

USD Lows — The US dollar index fell to a 3-year low as the euro and sterling hit highest since 2021, the Swiss franc hits a decade high.

Tax Bill Tensions — Senate Republicans are divided on Trump's $3.3T tax-cut bill. It is revised to reduce CFPB funding cuts to 6.5%. Musk again slams the bill as "utterly insane and destructive," predicting harm to jobs and the GOP.

AI Provision Rewrite — The Senate has rewritten the tax bill’s AI provision limiting state AI regulation down to a 5 year period from the 10 year limit in the originally tabled form.

Apple's Box Office Victory Lap — Apple's film "F1" races to success, earning $55.6 million domestically. Despite a hefty $200 million budget, it highlights Apple's innovative tech and marketing, driven by the popularity of Formula One.

IN PARTNERSHIP WITH DATASTAX

⚡ Forge the AI Frontier on NVIDIA Blackwell GPUs

Don’t fall behind in the rapidly evolving world of AI.

From our conversations with Fortune 500 leaders, we find out what are the keys to success on how to leverage enterprise and vector data for AI agents and applications.

Download this ebook to dig into each of these critical areas and get a clearer picture of what it takes to succeed with AI.

Meta's $65B Superintelligence Bet & Poaching Spree

Infrastructure, talent poaching, and the quiet rise of embodied AGI systems

Meta’s pursuit of superintelligence is taking shape. Mark Zuckerberg just announced the newly formed Meta Superintelligence Labs, an AI unit led by Alexandr Wang and legendary tech adviser Nat Friedman. The lab consolidates Meta’s sprawling AI initiatives of its LLaMA models, AR systems, smart glasses, and agents—under a single directive. The mission is unambiguous: build AI that exceeds human cognitive performance, across domains. Wang, former CEO of data-labeling startup Scale AI, will serve as chief AI officer. Meta recently aqui-hired him and top Scale talent as part of a drive to attract top AI talent it from startups and rivals like OpenAI. Zuckerberg has reportedly called Wang the “most impressive founder of his generation.”

Money Talks

This AGI mission is far from just talk. It is already becoming a capital project. In June, Meta began raising $29 billion in private financing to fund a network of AI-optimized data centers across the US. The move is part of a broader $65 billion AI investment strategy, placing it in direct competition with OpenAI and Google in both the model and the raw infrastructure battle.

The financing package includes $3B in equity and $26B in debt, backed by institutional partners like Apollo, Brookfield, and Carlyle. These investors are noted infrastructure capitalists who usually fund ports, power grids, and rail systems. That they are now underwriting AI data centers signals a deeper shift towards intelligence being treated as a utility.

Infrastructure First, Then AGI

Both the focus on talent and infrastructure is evidence Meta understands the Superintelligence race is about more than benchmarking battles. In 2025, what matters more than parameter count is runtime capacity. The ability to train, serve, and deploy across global latency zones now defines competitive edge. Elon Musk's xAI has been laser focused on this reality from inception. Google is onto it too, it just signed a deal to purchase 200 megawatts from Commonwealth Fusion Systems' Arc plant, operational by early 2030s, to support its expanding AI data centers.

Meta’s $29B raise is also designed to solve for this. Their new data centers will integrate proprietary chips, renewable energy sources, and direct-to-device distribution pipelines. Compute is the ground on which AGI-scale models will be built.

Talent flows toward the compute

This infrastructure-first strategy is accompanied by a calculated recruitment offensive. Over the past month, Meta has hired eight senior researchers from OpenAI, including contributors to GPT-4.1 and OpenAI’s multimodal models. Meta executives reportedly maintain a private “Recruiting Party” chat, where they share candidate papers and discuss incentives.

The flow of departures has prompted OpenAI to recalibrate internal compensation, pause internal deadlines, and circulate reassurances to staff. While Sam Altman brushed off the impact publicly, a deleted tweet from a researcher and a leaked internal memo suggest the opposite.

Researchers with experience in multimodal learning, efficient inference, and vision-language grounding are now high-priority hires. Meta’s headcount strategy reflects this, as do its product directions. The addition this week of Play AI, a voice cloning startup, and the release of Seamless Interaction—Meta’s full body avatar system that mirrors human gestures — show a coordinated pivot toward socially aware, embodied AI agents.

xAI is also playing the talent game, it just scooped up Nikita Bier, co-founder of social media app tbh, as its new head of product.

Multimodal Agents Are The Main Event 

Seamless Interaction was trained on more than 4,000 hours of dyadic conversation data. The model doesn’t merely respond to speech — it gestures, emotes, and synchronizes facial movement with language context. This type of behavioral realism is foundational to Meta’s smart glasses and AR assistant initiatives, but it also points toward a new kind of interface.

These agents aren’t designed to exist inside chat windows, they’re being built for wearables, cameras, games, and live interaction. The training pipelines are shaped by embodiment, not abstraction. It’s an engineering priority that differs from traditional NLP labs.

OpenAI’s latest acquisition this week of Crossing Minds — a company specializing in real-time recommendation personalization — suggests a parallel, albeit less embodied, move toward contextual adaptation. Where Meta is focused on social realism, OpenAI appears focused on optimizing interaction relevance. Both directions emerge from the same pressure: how to move beyond LLMs towards contextual agility. Anthropic's smart fridge fail revealed yesterday is an indication of the limitations of LLMs lack of contextual awareness. 

Google is also taking a more embodied track, with its latest DeepMind robotics update announced last week, demonstrating it understands the primacy of context.

China's Open-Source Model Explosion

Meanwhile, like Meta, China’s leading tech firms this week are releasing SOTA models under open-source licenses. Baidu open-sourced Ernie 4.5, a 424B parameter multimodal system with a mixture-of-experts architecture. While Tencent released Hunyuan-A13B an O-S reasoning model, and Alibaba dropped Qwen-VLo, an open source model tuned for creative and code-intensive tasks.

These releases were part of an industrial strategy supported by over $8.2 billion in state funds, as detailed in RAND’s recent AI policy analysis. By open-sourcing capable models, China is turning diffusion into leverage—enabling global adoption of Chinese models without requiring Chinese platforms.

The performance gap between proprietary and open models is narrowing. What remains is the difference in deployment infrastructure and alignment strategies — areas where the US models still hold an edge. But the economics are shifting. O-S models are 3.5× cheaper to deploy and now power nearly 2/3 of enterprise AI pilots worldwide.

China is now even coming for Meta's multimodal hardware. Xiaomi just launched its first AI-powered smart glasses, featuring a 12 MP camera, voice command capabilities, and an 8.6-hour battery life (double Meta's capacity). The starting price is just $278.54.

Superintelligent Metaverses

Meta’s superintelligence bet is far from being a side hustle, it is the main event. To fulfil its AGI ambitions it is doubling down on investment  in infrastructure for advanced models to thrive in and in the talent to build them. Its vision of the future is shaping up, and it's one characterized by multimodal socially interactive metaverse environments full of embodied superintelligent AI agents and real humans interacting. 

COMPANIES TO WATCH

Crossing Minds — The AI recommendation systems startup for e-commerce joins OpenAI, halting new client acquisitions to support OpenAI's mission in AGI.

Commonwealth Fusion Systems — Signed deal with Google to provide 200 mw from its Arc plant, operational by early 2030s, supporting its’ R&D and Google's energy needs for data centers.

TikTok — Trump announced a group of wealthy individuals is set to acquire the social media app pending Chinese approval, with buyer identities to be revealed in two weeks.

Glean — The Enterprise Search and data company’s success has caused major industry disruption, prompting big AI companies like OpenAI to develop rival products and restrict Glean's access to their corporate data.

Audos — Henrik Werdelin’s startup platform aims to launch 100,000 AI-driven companies annually, taking a 15% revenue share while providing tools to entrepreneurs not suited for VC.

Campfire — The AI-powered accounting startup raised a $35M Series A led by Accel to challenge legacy enterprise resource accounting software like NetSuite.

SignalPulse — Read the Top signals, Startups + Deals From The Last Week

TECH BUZZ DIRECTORY
Featured Products*

SEO Bot — A fully autonomous "SEO Robot" with AI agents for busy founders.

Memelord — Meme Software for marketing memes, tech memes, & sales memes.

Otonomos — Incorporate your Delaware C-corp or tax advantaged foundation and get 5% OFF.

BUZZY TOOLS

Buzzy Tech Tools To Watch & Use

OpenAI — New O-S model expected to surpass competitors, hinting at GPT-5.

Android 16 — Alerts against fake networks and tracking to boost security.

xAI's Grok 4 — July 4 launch w/ 1M Nvidia chips for enhanced training, adaptability.

Cursor — New web app helps users manage coding agents, view progress.

Xiaomi — Smart glasses with 12 MP cam, voice commands, 8.6-hr battery.

Spotify — Discover Weekly update lets Premium users personalize by genre.

Baidu — Ernie AI open-sourced to lower costs, challenge proprietary models.

BUZZY TECH

The Latest Deep Technology & Trends To Watch

Neuralink Mind Control — Patients play Call of Duty, control robotic hands.

5-D Physics — Model predicts mercury fission, revealing asymmetric behavior.

MAI Diagnostic — AI diagnoses diseases 4x more accurately than doctors.

Autonomous Soccer — Robots play first game; struggle with kicking, strategy.

Voice AI — Writers urge publishers to favor human narrators over AI.

CRYPTO WATCH

Robinhood — Launching native blockchain with 200 tokenized U.S. stocks expanding to 2,000 by year-end, plus crypto perpetual futures for European clients to democratize finance, according to CEO Vlad Tenev.

EU Crypto Rules — European lawmakers warn EU's lenient stablecoin regulations could trigger financial crisis and bank run, undermining Euro, concerning the ECB.

Bank of Korea — Halted its CBDC project to follow the U.S. lead and pivot instead to passing comprehensive stablecoin regulation.

Grayscale ETF — US SEC to decide by July 2 on converting Grayscale's Digital Large Cap Fund to spot ETF, per Bloomberg analysts.

PumpFun — Released iOS/Android mobile app with live streams and news for memecoin traders, supporting upcoming $PUMP token expected to raise $1B at $4B.

SPACE_RACE

Space Force_Maneuverable satellites + cloud software for orbital defense.

Blue Origin_Successful crewed suborbital flight, advances space tourism.

Shijian-21 Satellites_China refueling tests to reduce debris, extend life.

James Webb_Bullet Cluster images reveal galaxies, measure dark matter.

FOR INVESTORS
Open Deals
Investors– request an intro to startups - [email protected].

Before we restart our placements of raising startups, we’re sharing an open call to send in your own company. Just shoot us an email with a deck and we may feature you in this bulletin.

For detailed pitch materials, please email [email protected] with your deck.
🔥 — hot deal!
⏰ — leaving soon.

FOR EMPLOYERS
Job Candidates

Employers– request an intro to candidates - jobs@techbuzz.ai